Re: I/O blocked after booting

["Massimo B." <massimo.b@gmx.net>]

On Thu, 2024-03-28 at 20:40 +1030, Qu Wenruo wrote:
> Disable disable (nodiscard mount option), as it looks like there is
> something wrong with the auto discard, then retry.
> 
> This is mostly related to your NVME device's discard implementation, and
> I believe manually fstrim may be a better and more reliable solution.

Ok, I'm going to try.
Beside that I'm using different NVMe models and also an SATA SSD "Samsung SSD
860" and all are affected by that issue.

So you mean a daily cronjob for fstrim would be preferable?

Trying  fstrim -a -v  here with the very common Samsung SSD 970 PRO 1TB NVMe,
the command started discarding small partitions:

/boot/efi: 31,6 MiB (33183744 bytes) trimmed on /dev/nvme1n1p1
/boot: 0 B (0 bytes) trimmed on /dev/nvme1n1p2

but then does not return after 20 minutes. There is no IO on the disk which
might be ok for fstrim only communicating with the controler. But fstrim process
has some CPU usage...

And I see the same syslog entries after starting the fstrim:

Mar 28 11:38:26 [kernel] [14826.740669] BTRFS warning (device dm-0): failed to trim 698 block group(s), last error -512
Mar 28 11:38:26 [kernel] [14826.741731] BTRFS warning (device dm-0): failed to trim 1 device(s), last error -512

Best regards,
Massimo